Programmed cell death protein 1/programmed cell death protein ligand1 (PD-1/PD-L1) interaction is an important immune checkpoint targeted by anti-PD-1/PD-L1 immunotherapies. However, the observed prognostic significance of PD-1/PD-L1 expression in diffuse large B-cell lymphoma treated with the standard of care has been inconsistent and even contradictory. To clarify the prognostic role of PD-1/PD-L1 expression and interaction in diffuse large B-cell lymphoma, in this study we used 3-marker fluorescent multiplex immunohistochemistry and Automated Quantitative Analysis Technology to assess the CD3, PD-L1, and PD-1CD3 expression in diagnostic samples and PD-1/PD-L1 interaction as indicated by presence of PD-1CD3 cells in the vicinity of PD-L1 cells, analyzed their prognostic effects in 414 patients with de novo diffuse large B-cell lymphoma, and examined whether PD-1/PD-L1 interaction is required for the prognostic role of PD-1/PD-L1 expression. We found that low T-cell tissue cellularity, tissue PD-L1 expression (irrespective of cell types), PD-1CD3 expression, and PD-1/PD-L1 interaction showed hierarchical adverse prognostic effects in the study cohort. PD-1/PD-L1 interaction showed higher sensitivity and specificity than PD-1 and PD-L1 expression in predicting inferior prognosis in patients with high CD3 tissue cellularity ("hot"/inflammatory tumors). However, both PD-1 and PD-L1 expression showed adverse prognostic effects independent of PD-1/PD-L1 interaction, and PD-1/PD-L1 interaction showed favorable prognostic effect in PD-L1 patients without high CD3 tissue cellularity. Macrophage function and tumor-cell MYC expression may contribute to the PD-1-independent adverse prognostic effect of PD-L1 expression. In summary, low T-cell tissue cellularity has unfavorable prognostic impact in diffuse large B-cell lymphoma, and tissue PD-L1 expression and T-cell-derived PD-1 expression have significant adverse impact only in patients with high T-cell infiltration. PD-1/PD-L1 interaction in tissue is essential but not always responsible for the inhibitory effect of PD-L1/PD-1 expression. These results suggest the benefit of PD-1/PD-L1 blockade therapies only in patients with sufficient T-cell infiltration, and the potential of immunofluorescent assays and Automated Quantitative Analysis in the clinical assessment of PD-1/PD-L1 expression and interaction.